Santander, BBVA, CaixaBank y Sabadell exigen resultados a la IA

During recent years, Spanish banks have competed to announce agreements with technology giants, roll out copilots for employees and test dozens of AI projects. Now, the conversation is shifting. After an experimental phase focused on experimentation, entities are concentrating on scaling AI, transforming complete processes and achieving tangible business results.

Martí Riba, global leader of retail banking AI at McKinsey, summarizes this new stage as "finally fulfilling the promises of AI, and the pressure begins to shift from pilots to results." Banks claim they already perceive benefits, though most do not publicly quantify the potential impact on their profit and loss statements. Santander has already forecasted generating more than €1 billion in business value from AI between 2026 and 2028, with the first €200 million by the end of this fiscal year.

Dual path AI can contribute to business in two ways: reducing costs and boosting revenues. So far, most benefits have come from productivity and efficiency improvements, but the real breakthrough will come when AI is used to better understand customers, personalize offers and increase cross-selling of products and services. AI in banking has primarily focused on automating tasks, reducing times and cutting costs.

While this is a legitimate goal, it is insufficient because the capabilities that seem unique today will be standard tomorrow. Experts believe the next leap is transforming these capabilities into more revenue and better customer relationships, something that has not yet consolidated. Experts believe AI can still generate efficiency gains in an industry that already operates at high efficiency levels, with an average aggregate efficiency ratio of 41.12% in the six listed banks during the first quarter.

The search for these results is forcing banks to rethink their strategy. One of the most visible changes is moving beyond the "AI for all" model, based on deploying assistants and tools at scale to the workforce, which, according to Riba, has had a relatively low impact. The focus is on areas where AI can fundamentally change the way work is done.

It matters not how many agents you have or how many pilots you launch. What matters is how many domains you have transformed. Santander is working along these lines, with Ricardo Martín, head of Data and AI at the entity, stating that the focus should be on fewer initiatives, but with the capacity to move the needle, measure impact and scale what works across the group.

Sabadell is preparing a new stage, after deploying more than 50 use cases in 2026, the bank will evolve toward a transformation-based model, concentrating investment and AI capabilities in transformation domains to maximize impact. The new era demands industrializing AI capabilities. BBVA has launched a platform to create, deploy and manage thousands of intelligent agents quickly.

The bank is convinced that what will make the difference is being able to do this at scale, reusing developments without starting from scratch each time. The change in stage is evident at Bankinter. After making Microsoft Copilot available to the entire workforce and allowing employees to develop more than 7,000 agents, the entity has started a process to select tools with real potential to extend across the organization.

No longer is the goal developing the most agents, but having the best. Bankinter's director of Finance and Digital Banking, Jacobo Díaz, explains that the focus is now on selecting the best tools. Unicaja also considers the experiment phase over and is working on deploying AI at scale in the commercial business, where salespeople use conversational assistants to prepare offers and access product information; automating internal processes, from claims to document analysis; and developing software.

Unicaja says the use cases have allowed response times to be reduced by 40% to 80%. Where is AI being applied? Banks are using AI to develop software, automate internal processes, strengthen fraud and cybersecurity detection, improve commercial productivity, and deploy customer and employee assistants. One of the areas where implementation seems most advanced is software development.

Santander reports that nearly 40% of its technological teams' code already has AI support. BBVA highlights savings of 50% in development and testing time and a 28% increase in productivity. In Sabadell, 70% of these professionals already use AI, with productivity improvements of 20%. Bankinter is working to incorporate agents into the entire software development cycle.

Banks want AI to begin translating into more business. The bet is to use this technology to better understand customers, anticipate their needs and sell more. Intelligent assistants, for example, are being developed to answer queries, accompany customers during product contracts, offer personalized recommendations and execute operations.

CaixaBank, for example, has deployed a generative AI agent as the first point of contact on its web chats and mobile app.